The subjects are high school students who can follow instructions and are likely to have a high capacity for brain plasticity, which is important for experiments based on learning activities. Individuals with psychological disorders, blindness or visual impairments, and those unable to use the experimental materials were excluded as criteria.
The participants come from different places in Mexico, especially Durango, Sinaloa, Mexico City, and the State of Mexico. The sample consisted of 268 individuals (M age = 17.33, SD = 1.059), 158 (59%) females, and 109 (40.7%) males (Figure 12). Participants were asked for their informed consent, which involved clarifying the privacy of their data, stating their participation was voluntary, and explaining the purpose of the research.
Based on the primary objective of the research, a hypothesis was formulated for resolution, which justifies the calculation of the sample size (Quispe, 2020):
H0 = There are no statistically significant differences between the mean of accuracy in verbal concept learning and the mean of accuracy in nonverbal concept learning in the experiment.
The sampling was probabilistic, to calculate the sample size, RStudio software and the 'pwr' package titled "Basic Functions for Power Analysis" (Champely et al., 2017) were used. This package contains functions for Cohen's statistical power analysis (1988).
The hypothesis involves two means from different populations with two tails, i.e., H0: µ2 = µ1 vs HA: µ2 ≠ µ1.
To calculate the sample size, the effect size "d" or design effect needs to be obtained using the formula: (µ1 - µ2) / standard deviation. With µ1: 56, µ2: 53.43, and standard deviation: 7.024, the result is d = 0.36588838268792710706150341685649. The command for calculating the sample size for this type of hypothesis in RStudio is as follows:
Sample = pwr.t.test(d = 0.36588838268792710706150341685649, sig.level = 0.05, power = 0.50, type = 'two.sample', alternative = 'two.sided')
This resulted in a requirement of 58 subjects per group.
The subjects were matched based on their baseline anxiety, which represents the possibility of a person reacting with anxiety to life situations. Baseline anxiety is a variable that can affect learning (Blumer & Benson, 1975). To reduce confounding factors, the decision was made to match subjects based on this criterion. As a result of the matching process, three groups were obtained: high anxiety, medium anxiety, and low anxiety.
Each group obtained was randomly divided by the software into an experimental group and a control group (Figure 2), where the independent variable is verbal or nonverbal information.
At the end of the first experiment, the two low-anxiety groups terminated their participation in the trial since in the next experiment where anxiety is induced with a stressor, it is likely that the difference will be null. Therefore, these two groups were considered closed. The evaluation is performed automatically by the software at the moment that the subject interacts with it.
Data collection was conducted by the software application used in the experiment. For the assessment of subjects regarding the anxiety variable, the ASI-3 questionnaire was used: New Scale for the Assessment of Anxiety Sensitivity (Hernández-Pozo, Alvarado-Bravo, Espinosa-Luna, et al., 2022). The questionnaire was validated in the Mexican population with a total scale reliability of Cronbach's alpha equal to 0.919.
The data collected by the software were automatically sent to a spreadsheet on the network that captured data from all applications. This approach eliminated errors caused by the researcher's expectations during the experiment evaluation or personal observation.
Regarding the reliability of the experiment, with 268 valid cases and zero exclusions, 40 items were analyzed, resulting in a reliability coefficient of 0.885.
Experiment
An experiment was designed that consisted of two stages. In the first stage, the subjects were repeatedly exposed to a set of related stimuli to create a concept, as proposed by the engram and informational atomism theories when referring to the phenomenon known as "locking". The nature of the stimuli changed according to a verbal and nonverbal taxonomy, which defined the control and experimental groups.
In the second stage, concept learning was tested. During this part, a stimulus was presented to the subjects, and they were asked to relate it to the properties they had associated with it. This process aimed to recreate the theories that frame this research.
Since the design was double-blind, the response to the hypothesis could be determined at the end of the experiment because the subjects were randomly assigned to either the control or experimental group, without their knowledge of the subjects or the researcher, of which group they belonged to.
The experiments were double-balanced in terms of learning and probing. In the training or learning phase, the training was balanced based on concepts. For each concept, seven exposures were conducted, resulting in a total of four concepts.
Table 2 Distribution example
Trial
|
Discriminative Stimuli
|
|
Concepto 1
|
Concepto 2
|
1
|
2
|
1
|
2
|
0
|
4
|
3
|
1
|
2
|
4
|
4
|
0
|
5
|
3
|
3
|
For Experiment 1, there were 5 probing trials per concept. Each trial consisted of a sample stimulus, six comparison stimuli, and discriminative stimuli presented with a balanced distribution of correct responses ranging from 0 to 4 for each concept (Table 2).
Experiment 2 consisted of concept retrieval exercises or information blocks. It comprised 10 probing trials, in which there were 4 possible responses, with only one being correct.